Abstract:In order to support the demand of real-time transaction and safety check in the spot electricity market, a short-term load interval prediction method based on classified volatility statistics is proposed. Firstly, the traditional statistical analysis of load volatility and the calculation method of interval prediction limit are introduced. By standardizating the historical data of load, the distribution histogram of load volatility is drawn, and the interval limit under the given precision is calculated. Then, based on the reality of China, this paper discusses the fluctuation characteristics of load of different types of users from the perspectives of residents, commercial and industrial users, and summarizes the upper and lower limits of load interval prediction of the whole network, so as to realize the interval prediction of the whole network load. Finally, a case study constructed based on the actual data of a certain province shows that through in-depth analysis of the fluctuation of different types of loads, the method proposed in this paper can achieve the overall optimal prediction accuracy and interval width, reduce the interval width and improve the practicability of load interval prediction results on the premise of ensuring the same prediction accuracy.